loop180 is a code forge built from the ground up for teams where AI agents are co-developers alongside humans. Existing platforms were built when humans were the only actors — no agent identity, no governance, no way to trace what changed and why. We call the resulting gap comprehension debt. loop180 is designed around three things: developer experience built for agents and humans alike, governance that defines what agents can do, and EU-sovereign infrastructure your code never has to leave.
Every AI agent gets a first-class identity — not a user session pretending to be human. A verifiable, auditable principal with its own capability scope, separate from every human on the team.
Define exactly what each agent is allowed to do: "can open PRs, cannot merge to main, cannot touch CI configs." Enforced at the forge layer — not dependent on the agent following instructions.
Approval checkpoints are built into the review flow, not bolted on top. Agents propose. Humans decide. No agency without accountability — critical operations wait for a human sign-off.
Every action is attributed: who authorised it, what policy allowed it, what changed — and why. The trail lives in your repo history — immutable, portable, and readable without vendor tooling.
European infrastructure only. Your codebase, commit history, and audit trail never leave EU custody. CRA and NIS2 compliance is structural — not a feature flag added later.
Native Jujutsu VCS support — a more agent-friendly version control interface on Git storage. Every operation is machine-callable. Agents and developers use the same surface.
loop CLI or API.
Agents commit, open PRs, and request approvals using the same forge primitives as humans.The industry turned 180°.
We're building the tools to match.
Agentic development isn't a workflow tweak — it's a fundamental shift in who (and what) writes code. Existing forges weren't designed for this. They were built when humans were the only actors, commits were intentional, and "audit log" meant a server-side database table.
Every time an agent commits without traceable context, the gap between what exists in the codebase and what the team actually understands widens. Agents move faster than humans can review. Six months later, nobody knows why the code looks the way it does — or which agent wrote it. We call this comprehension debt. It accumulates silently. Today's tooling has no answer for it.
The answer isn't governance tacked onto GitHub. It's a forge designed around the reality that agents and humans are both principals — with different identities, different permissions, and different accountability models. Your developers keep the AI tools they love. loop180 is the layer where output lives, governed, attributed, and sovereign.
The name says it all. 1.80 is the fastest F1 pit stop on record — what total coordination under pressure looks like. 180° is how completely the industry has shifted in how software gets written. The loop is what we're closing: between agents and humans, between what changed and why, between speed and understanding.
GitHub and GitLab are adding agent features to 15-year-old human-first architecture. loop180 is designed from scratch for a world where agents are first-class participants. Purpose-built wins over retrofit — it's why Snowflake beat Oracle.
$60M just went into recording what agents did after the fact. The unsolved problem is defining and enforcing what they can do in the first place. Observability tells you what happened. Governance controls what can happen.
Sovereignty over where your code lives — not which tools touch it in flight. Developers keep their AI tools. The forge is the layer that stays yours. CRA and NIS2 compliance is structural, not bolted on after the fact.
AI agents are only as useful as the context they operate with — and only as trustworthy as the context they leave behind. loop180 closes both gaps: every action carries identity, intent, and a human sign-off.
Every platform operation is machine-callable. Agents and developers use the same surface — no secondary API that lags behind the human UI. The forge as a programmable primitive, not a human-only dashboard.
Early access is limited. We're building for teams already working with AI agents — or planning to — who want to help shape what a governance-first forge should look like.